| Optimization problems widely exist in industry, agriculture, national defense, engineering,transportation, finance, chemicals, energy, communications, IT and many other fields.Practice shows that under the same conditions, systems processed by optimization techniqueswill improve energy efficiency, reduce energy consumption, allocate resources reasonably,and helpful to the improvement of economic efficiency. With the rapid development ofmodern production technology, the optimization calculation of the various types ofengineering problems are more and more complex, the calculation accuracy requirements arealso increasing. Traditional optimization methods which based on rigorous mechanism modelare unable to meet the actual demand of production activities.The thesis is based on the principle of artificial intelligence originated from biologicalbehaviors is used for the solution of optimization problems via biological method, then anarchitecture of problem solving, which we called artificial root method, will be defined, andan efficient intelligent optimization algorithm will be introduced as artificial root algorithm(AR).The principle and description of AR are described in detail. The convergence property ofthe algorithm was discussed by the classic optimization test function; artificial rootoptimization algorithm is applied to estimate the aquifer parameters by analyzing the date ofpumping test, and obtain the reliable calculations; a new type of discriminant analysismethods are proposed, and apply them to identify the type of fluid unit having similarsedimentary and flow characteristics; at last, the future development of AR is proposed.Some characters of AR are as follows:Algorithm has fast optimization ability to quickly converge to the global optimalvalue;Algorithm has the heuristic search ability using the method of combining local searchand global search;Algorithm is stable, and can effectively find the global optimal solution to find theglobal optimum nonlinear multi-modal function;Algorithm has the characteristics of parallelism, the artificial root relative independently and to search in parallel;Algorithm has global immunity for local extreme, and a strong ability to jump out oflocal minima.The artificial method and AR are different from traditional methods, and they have thefoundations of merging with traditional approaches, it is easy to understand and realized, webelieve that AR can be widely applied in the future. |
